CN116167989A - Intelligent production method and system for aluminum cup - Google Patents

Intelligent production method and system for aluminum cup Download PDF

Info

Publication number
CN116167989A
CN116167989A CN202310097580.XA CN202310097580A CN116167989A CN 116167989 A CN116167989 A CN 116167989A CN 202310097580 A CN202310097580 A CN 202310097580A CN 116167989 A CN116167989 A CN 116167989A
Authority
CN
China
Prior art keywords
training
classification
aluminum cup
detection image
classifier
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
CN202310097580.XA
Other languages
Chinese (zh)
Inventor
陈剑永
罗启甲
陈程
杨晓
林天郎
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shengxing Bode New Material Wenzhou Co ltd
Original Assignee
Shengxing Bode New Material Wenzhou Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shengxing Bode New Material Wenzhou Co ltd filed Critical Shengxing Bode New Material Wenzhou Co ltd
Priority to CN202310097580.XA priority Critical patent/CN116167989A/en
Publication of CN116167989A publication Critical patent/CN116167989A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Quality & Reliability (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

The intelligent production method and the system thereof for the aluminum cup adopt an artificial intelligent detection technology based on deep learning to perform characteristic excavation of an interested region of the hole by performing target anchoring on the position of the hole in a detection image so as to extract hidden characteristic information about the hole with a small scale, and further utilize the position characteristic information about the hole in the detection image so as to comprehensively detect whether the punching of the machined and formed aluminum cup is deviated. Therefore, the punching quality of the machined aluminum cup can be accurately detected, and the working efficiency and the production quality of the machining of the aluminum cup are improved.

Description

Intelligent production method and system for aluminum cup
Technical Field
The application relates to the technical field of aluminum cup production, and more particularly relates to an intelligent production method and system of an aluminum cup.
Background
Aluminum cups are used in vessels because they have a dense oxide film on their surface which is not easily corroded. Many enterprises are producing aluminum cups at present, but when the shape and structure of the aluminum cups are complex, the preparation process of the aluminum cups is challenged.
When a general enterprise processes an aluminum cup, the processing and forming process comprises the following steps: the worker takes the aluminum alloy material to a lathe to carry out turning forming, and the formed aluminum cup also needs to be taken out of the drilling machine to carry out drilling. That is, different processing steps are performed on different processing equipment, and the mode has low working efficiency and complex processing steps, and safety accidents can also occur in manual operation.
Aiming at the technical problems, chinese patent CN 103691806B discloses a process for processing and forming an aluminum cup, which comprises the following steps: first, one-time preforming processing of an aluminum cup: placing an aluminum alloy material to be processed on preforming processing equipment, matching an upper die and a lower die in the preforming processing equipment, and stamping and forming the aluminum alloy material; secondly, positioning and punching the aluminum cup: and placing the processed and formed semi-finished aluminum cup on a positioning and punching device, and punching the aluminum cup by a punch on an upper die in the positioning and punching device.
Although the process for forming the aluminum cup can optimize the working efficiency, the punching offset caused by vibration or equipment displacement in the punching stage is found in the production process, and a large number of defective products are found in time.
Thus, a more optimal intelligent production scheme for aluminum cups is desired.
Disclosure of Invention
The present application has been made in order to solve the above technical problems. The embodiment of the application provides an intelligent production method and system of an aluminum cup, which adopt an artificial intelligent detection technology based on deep learning to perform feature mining of an interesting region of a hole by anchoring a target at the position of the hole in a detection image, so as to extract hidden feature information about the hole with a small scale, and also use the position feature information about the hole in the detection image to comprehensively detect whether the punching of the machined aluminum cup is deviated or not. Therefore, the punching quality of the machined aluminum cup can be accurately detected, and the working efficiency and the production quality of the machining of the aluminum cup are improved.
According to one aspect of the present application, there is provided an intelligent production method of an aluminum cup, comprising: acquiring a detection image of the machined and formed aluminum cup; applying a mask to the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a mask image; aggregating the mask image and the detection image along a channel dimension to obtain a multi-channel detection image; the multi-channel detection image is subjected to a convolutional neural network model using a spatial attention mechanism to obtain a classification characteristic diagram; and the classification characteristic diagram is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the punching of the machined and formed aluminum cup is deviated or not.
In the above intelligent production method of an aluminum cup, the masking of the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a masking image includes: and enabling the detection image to pass through a hole target identification network so as to anchor the position of the hole in the aluminum cup.
In the intelligent production method of the aluminum cup, the hole target recognition network is Fast R-CNN or Fast R-CNN, retinaNet.
In the above-mentioned intelligent production method of aluminum cup, the step of obtaining the classification feature map by using a convolutional neural network model of a spatial attention mechanism from the multi-channel detection image includes: performing depth convolution encoding on the multi-channel detection image by using a convolution encoding part of the convolution neural network model to obtain an initial convolution characteristic diagram; inputting the initial convolution feature map into a spatial attention portion of the convolution neural network model to obtain a spatial attention map; -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile; and calculating the position-wise point multiplication of the spatial attention characteristic diagram and the initial convolution characteristic diagram to obtain a classification characteristic diagram.
In the above intelligent production method of aluminum cup, the classifying feature map is passed through a classifier to obtain a classifying result, where the classifying result is used to indicate whether the punching of the machined aluminum cup is offset, and the method includes: expanding each classification feature matrix in the classification feature map into a one-dimensional feature vector according to a row vector or a column vector, and then cascading to obtain a classification feature vector; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
The intelligent production method of the aluminum cup further comprises the training steps of: training the convolutional neural network model using a spatial attention mechanism and the classifier.
In the above intelligent production method of aluminum cup, the training step includes: acquiring training data, wherein the training data comprises training detection images of the machined and formed aluminum cup and a true value of whether the punching of the machined and formed aluminum cup is deviated or not; applying a mask to the corresponding position of the training detection image based on the position of the hole in the aluminum cup to obtain a training mask image; aggregating the training mask image and the training detection image along a channel dimension to obtain a training multi-channel detection image; passing the training multichannel detection image through the convolutional neural network model using a spatial attention mechanism to obtain a training classification characteristic diagram; passing the training classification feature map through the classifier to obtain a classification loss function value; and training the convolutional neural network model using a spatial attention mechanism and the classifier based on the classification loss function value and by back propagation of gradient descent, wherein, in each round of iteration of the training process, a free label optimization factor of the training classification feature map based on cross-classifier soft similarity is calculated as a label value of the classifier.
In the above-mentioned intelligent production method of aluminum cup, in each iteration of the training process, a free label optimization factor based on cross-classifier soft similarity of the training classification feature vector is calculated as a label value of the classifier according to the following formula; wherein, the formula is:
Figure SMS_1
，
wherein the method comprises the steps of
Figure SMS_3
Is the training classification feature vector obtained after the training classification feature map is unfolded, and is +.>
Figure SMS_7
Is a weight matrix of training classification feature vectors obtained by the classifier after the training classification feature map is developed, and is->
Figure SMS_9
And->
Figure SMS_4
Representing tensor multiplication and tensor addition, respectively,/->
Figure SMS_6
Representing the distance between vectors, < >>
Figure SMS_8
Representing the two norms of the vector, and +.>
Figure SMS_11
And->
Figure SMS_2
Is a weight superparameter,/->
Figure SMS_5
An exponential operation representing a vector representing the calculation of the self-power of the eigenvalues of each position in the vectorNatural index function value->
Figure SMS_10
A tag value representing the classifier.
According to another aspect of the present application, there is provided an intelligent production system for aluminum cups, comprising: the detection image acquisition unit is used for acquiring a detection image of the machined and molded aluminum cup; a mask applying unit for applying a mask to the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a mask image; a multi-channel aggregation unit, configured to aggregate the mask image and the detection image along a channel dimension to obtain a multi-channel detection image; a spatial attention applying unit for applying the multichannel detection image to a convolutional neural network model using a spatial attention mechanism to obtain a classification characteristic map; and the detection result generation unit is used for passing the classification characteristic diagram through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the punching of the machined and formed aluminum cup is deviated or not.
In the intelligent production system of the aluminum cup, the mask applying unit is used for enabling the detection image to pass through a hole target recognition network so as to anchor the position of the hole in the aluminum cup.
In the intelligent production system of the aluminum cup, the hole target identification network is Fast R-CNN or Fast R-CNN, retinaNet.
In the above-mentioned intelligent production system of aluminum cup, the spatial attention applying unit is further configured to: performing depth convolution encoding on the multi-channel detection image by using a convolution encoding part of the convolution neural network model to obtain an initial convolution characteristic diagram; inputting the initial convolution feature map into a spatial attention portion of the convolution neural network model to obtain a spatial attention map; -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile; and calculating the position-wise point multiplication of the spatial attention characteristic diagram and the initial convolution characteristic diagram to obtain a classification characteristic diagram.
In the above-mentioned intelligent production system of aluminum cup, the detection result generating unit is further configured to: expanding each classification feature matrix in the classification feature map into a one-dimensional feature vector according to a row vector or a column vector, and then cascading to obtain a classification feature vector; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
The intelligent production system of the aluminum cup further comprises a training module for training the convolutional neural network model using the spatial attention mechanism and the classifier.
In the intelligent production system of above-mentioned aluminum cup, training module includes: the training data acquisition unit is used for acquiring training data, wherein the training data comprises training detection images of the machined and formed aluminum cup and a true value of whether the punching of the machined and formed aluminum cup is deviated or not; a training mask applying unit, configured to apply a mask to the corresponding position of the training detection image based on the position of the hole in the aluminum cup to obtain a training mask image; the training multichannel aggregation unit is used for aggregating the training mask image and the training detection image along the channel dimension to obtain a training multichannel detection image; the training spatial attention applying unit is used for enabling the training multichannel detection image to pass through the convolutional neural network model using the spatial attention mechanism so as to obtain a training classification characteristic diagram; the classification loss unit is used for passing the training classification characteristic diagram through the classifier to obtain a classification loss function value; and a training unit for training the convolutional neural network model using a spatial attention mechanism and the classifier based on the classification loss function value and by back propagation of gradient descent, wherein, in each round of iteration of the training process, a free label optimization factor of the training classification feature map based on cross-classifier soft similarity is calculated as a label value of the classifier.
In the above-mentioned intelligent production system of aluminum cups, in each iteration of the training process, a free label optimization factor of the training classification feature vector based on cross-classifier soft similarity is calculated as a label value of the classifier according to the following formula; wherein, the formula is:
Figure SMS_12
，
wherein the method comprises the steps of
Figure SMS_15
Is the training classification feature vector obtained after the training classification feature map is unfolded, and is +.>
Figure SMS_18
Is a weight matrix of training classification feature vectors obtained by the classifier after the training classification feature map is developed, and is->
Figure SMS_21
And->
Figure SMS_14
Representing tensor multiplication and tensor addition, respectively,/->
Figure SMS_17
Representing the distance between vectors, < >>
Figure SMS_20
Representing the two norms of the vector, and +.>
Figure SMS_22
And->
Figure SMS_13
Is a weight superparameter,/->
Figure SMS_16
An exponential operation representing a vector representing a calculation of a natural exponential function value raised to a power by a characteristic value of each position in the vector, ">
Figure SMS_19
A tag value representing the classifier.
According to still another aspect of the present application, there is provided an electronic apparatus including: a processor; and a memory in which computer program instructions are stored which, when executed by the processor, cause the processor to perform the intelligent production method of aluminium cups as described above.
According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the method of intelligent production of aluminium cups as described above.
Compared with the prior art, the intelligent production method and the system thereof for the aluminum cup adopt an artificial intelligent detection technology based on deep learning, so that feature mining of the interesting region of the hole is carried out by carrying out target anchoring on the position of the hole in a detection image, thus the hidden feature information about the hole with a small scale is extracted, and the position feature information about the hole in the detection image is also utilized, so that whether the punching of the machined and molded aluminum cup is deviated or not is comprehensively detected. Therefore, the punching quality of the machined aluminum cup can be accurately detected, and the working efficiency and the production quality of the machining of the aluminum cup are improved.
Drawings
The foregoing and other objects, features and advantages of the present application will become more apparent from the following more particular description of embodiments of the present application, as illustrated in the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.
Fig. 1 is an application scenario diagram of an intelligent production method of an aluminum cup according to an embodiment of the application.
Fig. 2 is a flow chart of an intelligent production method of an aluminum cup according to an embodiment of the application.
Fig. 3 is a schematic diagram of an intelligent production method of aluminum cups according to an embodiment of the present application.
Fig. 4 is a flowchart for training the convolutional neural network model using the spatial attention mechanism and the classifier in the intelligent production method of aluminum cups according to the embodiment of the application.
Fig. 5 is a block diagram of an intelligent production system for aluminum cups according to an embodiment of the present application.
FIG. 6 is a block diagram of a training module in an intelligent production system for aluminum cups according to an embodiment of the application.
Fig. 7 is a block diagram of an electronic device according to an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Summary of the application: as described above, in a process of forming an aluminum cup disclosed in patent CN 103691806B, although the process of forming an aluminum cup can optimize the working efficiency, it is found that the punching offset is caused by vibration or displacement of equipment at the punching stage during the production process, and if it is not found in time, a large number of defective products are caused. Thus, a more optimal intelligent production scheme for aluminum cups is desired.
Specifically, in the actual production and manufacturing process of the aluminum cup, quality detection is required to be carried out on punching of the aluminum cup, so that the punching qualification rate of the aluminum cup and the processing and forming quality of the aluminum cup are ensured. Accordingly, it is considered that when the punching quality is detected, analysis can be performed on the detected image of the machined aluminum cup, but because the amount of information existing in the detected image of the machined aluminum cup is large, a large amount of useless information interference exists, and characteristic information about holes of the aluminum cup is a small-scale characteristic in the image, the holes are difficult to mine and acquire.
In recent years, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks have also shown levels approaching and even exceeding humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
The development of deep learning and neural networks provides a new solution idea and scheme for the perforation quality detection of the formed aluminum cup.
Specifically, in the technical scheme of the application, an artificial intelligent detection technology based on deep learning is adopted to perform feature mining of the hole interested region by anchoring a target at the position of the hole in a detection image, so as to extract hidden feature information about the hole with a small scale, and the position feature information about the hole in the detection image is also utilized to comprehensively detect whether the punching of the machined aluminum cup is deviated or not. Therefore, the punching quality of the machined aluminum cup can be accurately detected, and the working efficiency and the production quality of the machining of the aluminum cup are improved.
More specifically, in the technical scheme of the application, firstly, a detection image of a machined and molded aluminum cup is obtained through a camera. Next, considering that, in the detected image, whether the punching of the machined aluminum cup is shifted is detected in relation to the shape, size and smoothness of the hole, the feature information of the hole in the detected image is a feature of a small scale, and the feature information is affected by other useless disturbance features during feature extraction, so that it is difficult to perform implicit feature mining of the hole. Therefore, it is necessary to focus on the implicit features of the hole area during feature extraction and filter out the rest of useless interference feature information. Based on the above, in the technical scheme of the application, a mask is applied to the corresponding position of the detection image based on the position of the hole in the aluminum cup so as to obtain a mask image. Specifically, the detection image is passed through a hole target recognition network to anchor the holes at the position of the aluminum cup. In particular, here, the hole target recognition network is an anchor window based target recognition network, and the anchor window based target recognition network is Fast R-CNN, fast R-CNN or RetinaNet.
Then, considering that when the hole punching of the machined aluminum cup is offset to be detected, the position characteristic information of the hole in the detected image needs to be focused, in the technical scheme of the application, the mask image and the detected image are further aggregated along the channel dimension, so that a multi-channel detected image with the hole information and the original image information focused is obtained.
Further, the convolutional neural network model with excellent performance in terms of implicit feature extraction of the image is used for feature extraction of the multi-channel detection image, particularly, considering that implicit feature information and position feature information about holes on spatial positions are considered to be ignored for disturbing features irrelevant to the punching quality of the aluminum cup when punching quality detection of the machined aluminum cup is performed, focusing positions can be selected in view of an attention mechanism, a more resolved feature representation is generated, and the features added into an attention module can change adaptively with the deepening of the network. Therefore, in the technical scheme of the application, the multi-channel detection image is processed in a convolutional neural network model by using a spatial attention mechanism so as to extract hidden characteristic distribution information and spatial position distribution characteristic information focused on a space on holes in the detection image of the aluminum cup, thereby obtaining a classification characteristic diagram. It should be noted that, here, the image features extracted by the spatial attention reflect weights of differences of features of spatial dimensions, so as to suppress or strengthen features of different spatial positions, so as to extract feature information focused on the machined aluminum cup spatially.
And then, classifying the classification characteristic map with the implicit characteristic and the spatial position characteristic distribution information of the aluminum cup holes in a classifier to obtain a classification result for indicating whether the punching of the machined aluminum cup is deviated. That is, in the technical solution of the present application, the label of the classifier includes that the punching of the machined aluminum cup is offset, and that the punching of the machined aluminum cup is not offset, where the classifier determines, through a soft maximum function, to which classification label the classification feature map belongs. Therefore, the punching quality of the machined and formed aluminum cup can be detected, so that the working efficiency and the production quality of the machining of the aluminum cup are improved.
In particular, in the technical solution of the present application, the mask image and the detection image are aggregated along a channel dimension to obtain the multi-channel detection image, so that high-dimensional image semantic features of the mask image and the detection image can be fully extracted, but at the same time, expression of the high-dimensional image semantic features of the mask image and the detection image in the classification feature map can reduce association degree of feature distribution of the whole classification feature map. Further, since the multi-channel detection image obtains the classification feature map through a convolutional neural network model using a spatial attention mechanism, the association of the high-dimensional spatial semantic feature distribution between the mask image and the detection image is further weakened under the condition of strengthening the respective high-dimensional spatial semantic feature distribution of the mask image and the detection image, so that the overall feature distribution of the classification feature map may have strong discreteness, and training of a classifier, particularly convergence of a label value of the classifier is difficult.
Thus, soft label learning is preferably used instead of usual hard label learning, in particular, at each iteration, a free label optimization factor of the classification feature map based on cross-classifier soft similarity is calculated as the label value of the classifier, expressed as:
Figure SMS_23
，
Figure SMS_25
is the classification characteristic vector obtained after the expansion of the classification characteristic diagram,/for the classification characteristic vector>
Figure SMS_28
Is the classifier to theClassification feature vector->
Figure SMS_30
Weight matrix of>
Figure SMS_26
And->
Figure SMS_29
Representing tensor multiplication and tensor addition, respectively,/->
Figure SMS_31
Representing the distance between vectors, < >>
Figure SMS_32
Representing the two norms of the vector, and +.>
Figure SMS_24
And->
Figure SMS_27
Is a weight super parameter.
The free label optimization factor based on the cross-classifier soft similarity is used as a label value of the classifier, the classification probability of the classification feature map can be calculated instead of a hard label value, the classification feature vector obtained after the development of the classification feature map and the weight matrix of the classifier are subjected to bidirectional clustering, so that the pseudo class based on the weight matrix of the classifier is simulated through the soft similarity of the classification feature vector and the weight matrix, and therefore classification quantization loss caused by the hard label learning is avoided through soft similarity learning, free label optimization which is more focused on an internal weight structure of the classifier is realized, training of the label value of the classifier is optimized, and training speed of the classifier is improved. Therefore, the punching quality of the machined aluminum cup can be accurately detected, and the working efficiency and the production quality of the machining of the aluminum cup are improved.
Based on this, the application provides an intelligent production method of an aluminum cup, which comprises the following steps: acquiring a detection image of the machined and formed aluminum cup; applying a mask to the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a mask image; aggregating the mask image and the detection image along a channel dimension to obtain a multi-channel detection image; the multi-channel detection image is subjected to a convolutional neural network model using a spatial attention mechanism to obtain a classification characteristic diagram; and the classification characteristic diagram is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the punching of the machined and formed aluminum cup is deviated or not.
Fig. 1 is an application scenario diagram of an intelligent production method of an aluminum cup according to an embodiment of the application. As shown in fig. 1, in this application scenario, first, a detection image of a machined aluminum cup (e.g., cu as illustrated in fig. 1) is acquired by a camera (e.g., ca as illustrated in fig. 1). Further, the detected image of the formed aluminum cup is input to a server (e.g., S as illustrated in fig. 1) in which an intelligent production algorithm of the aluminum cup is deployed, wherein the server is capable of processing the detected image of the formed aluminum cup based on the intelligent production algorithm of the aluminum cup to obtain a classification result indicating whether or not punching of the formed aluminum cup is offset.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.
An exemplary method is: fig. 2 is a flow chart of an intelligent production method of an aluminum cup according to an embodiment of the application. As shown in fig. 2, the intelligent production method of the aluminum cup according to the embodiment of the application comprises the following steps: s110, obtaining a detection image of the machined and molded aluminum cup; s120, masking the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a masking image; s130, aggregating the mask image and the detection image along a channel dimension to obtain a multi-channel detection image; s140, the multichannel detection image is subjected to a convolutional neural network model using a spatial attention mechanism to obtain a classification characteristic diagram; and S150, passing the classification characteristic diagram through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the punching of the machined aluminum cup is deviated or not.
Fig. 3 is a schematic diagram of an intelligent production method of aluminum cups according to an embodiment of the present application. In this architecture, as shown in fig. 3, first, a detection image of a machined aluminum cup is acquired; then, masking the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a masking image; then, aggregating the mask image and the detection image along a channel dimension to obtain a multi-channel detection image; then, the multichannel detection image is subjected to a convolutional neural network model using a spatial attention mechanism to obtain a classification characteristic diagram; and finally, the classification characteristic diagram is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the punching of the machined and formed aluminum cup is deviated or not.
As described above, in a process of forming an aluminum cup disclosed in patent CN 103691806B, although the process of forming an aluminum cup can optimize the working efficiency, it is found that the punching offset is caused by vibration or displacement of equipment at the punching stage during the production process, and if it is not found in time, a large number of defective products are caused. Thus, a more optimal intelligent production scheme for aluminum cups is desired.
Specifically, in the actual production and manufacturing process of the aluminum cup, quality detection is required to be carried out on punching of the aluminum cup, so that the punching qualification rate of the aluminum cup and the processing and forming quality of the aluminum cup are ensured. Accordingly, it is considered that when the punching quality is detected, analysis can be performed on the detected image of the machined aluminum cup, but because the amount of information existing in the detected image of the machined aluminum cup is large, a large amount of useless information interference exists, and characteristic information about holes of the aluminum cup is a small-scale characteristic in the image, the holes are difficult to mine and acquire.
In recent years, deep learning and neural networks have been widely used in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks have also shown levels approaching and even exceeding humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
The development of deep learning and neural networks provides a new solution idea and scheme for the perforation quality detection of the formed aluminum cup.
Specifically, in the technical scheme of the application, an artificial intelligent detection technology based on deep learning is adopted to perform feature mining of the hole interested region by anchoring a target at the position of the hole in a detection image, so as to extract hidden feature information about the hole with a small scale, and the position feature information about the hole in the detection image is also utilized to comprehensively detect whether the punching of the machined aluminum cup is deviated or not. Therefore, the punching quality of the machined aluminum cup can be accurately detected, and the working efficiency and the production quality of the machining of the aluminum cup are improved.
In step S110, a detection image of the machined aluminum cup is acquired. In the technical scheme of the application, the detection image of the machined aluminum cup can be obtained through the camera.
In step S120, a mask is applied to the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a mask image. Considering that in the detection image, whether the punching of the machined aluminum cup is deviated or not is related to the shape, the size and the smoothness of the hole, and in the detection image, the characteristic information of the hole in the detection image is a small-scale characteristic, and the characteristic information is influenced by other useless interference characteristics during characteristic extraction, so that the hidden characteristic mining of the hole is difficult to carry out. Therefore, it is necessary to focus on the implicit features of the hole area during feature extraction and filter out the rest of useless interference feature information. Based on the above, in the technical scheme of the application, a mask is applied to the corresponding position of the detection image based on the position of the hole in the aluminum cup so as to obtain a mask image.
Specifically, the detection image is passed through a hole target recognition network to anchor the holes at the position of the aluminum cup. In particular, here, the hole target recognition network is an anchor window based target recognition network, and the anchor window based target recognition network is Fast R-CNN, fast R-CNN or RetinaNet.
In step S130, the mask image and the detection image are aggregated along a channel dimension to obtain a multi-channel detection image. In view of whether the hole is offset detected during the punching of the machined aluminum cup, the position feature information of the hole in the detected image needs to be focused, so in the technical scheme of the application, the mask image and the detected image are further aggregated along the channel dimension, and a multi-channel detected image with the hole information and the original image information is obtained.
In step S140, the multi-channel detection image is passed through a convolutional neural network model using a spatial attention mechanism to obtain a classification feature map. That is, the feature extraction of the multi-channel detected image is performed using a convolutional neural network model having excellent performance in terms of implicit feature extraction of the image, particularly, considering that implicit feature information and position feature information about holes in spatial positions should be focused and disturbance features irrelevant to the punching quality of the aluminum cup should be ignored when punching quality detection of the machined aluminum cup is performed, focusing positions can be selected in view of an attention mechanism, a more resolved feature representation is generated, and the features added to an attention module may change adaptively with deepening of the network. Therefore, in the technical scheme of the application, the multi-channel detection image is processed in a convolutional neural network model by using a spatial attention mechanism so as to extract hidden characteristic distribution information and spatial position distribution characteristic information focused on a space on holes in the detection image of the aluminum cup, thereby obtaining a classification characteristic diagram. It should be noted that, here, the image features extracted by the spatial attention reflect weights of differences of features of spatial dimensions, so as to suppress or strengthen features of different spatial positions, so as to extract feature information focused on the machined aluminum cup spatially.
Specifically, in an embodiment of the present application, the step of obtaining the classification feature map by using a convolutional neural network model of a spatial attention mechanism by using the multi-channel detection image includes: performing depth convolution encoding on the multi-channel detection image by using a convolution encoding part of the convolution neural network model to obtain an initial convolution characteristic diagram; inputting the initial convolution feature map into a spatial attention portion of the convolution neural network model to obtain a spatial attention map; -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile; and calculating the position-wise point multiplication of the spatial attention characteristic diagram and the initial convolution characteristic diagram to obtain a classification characteristic diagram.
In step S150, the classification feature map is passed through a classifier to obtain a classification result, where the classification result is used to indicate whether the punching of the machined aluminum cup is offset. That is, the classification feature map with the implicit feature and the spatial position feature distribution information of the aluminum cup holes is subjected to classification processing in the classifier to obtain a classification result for indicating whether the punching of the formed aluminum cup is deviated. In the technical scheme of the application, the label of the classifier comprises that the punching of the machined aluminum cup is deviated and that the punching of the machined aluminum cup is not deviated, wherein the classifier determines which classification label the classification feature map belongs to through a soft maximum function. Therefore, the punching quality of the machined and formed aluminum cup can be detected, so that the working efficiency and the production quality of the machining of the aluminum cup are improved.
Specifically, in an embodiment of the present application, the step of passing the classification feature map through a classifier to obtain a classification result, where the classification result is used to indicate whether the punching of the machined aluminum cup is offset, and the method includes: expanding each classification feature matrix in the classification feature map into a one-dimensional feature vector according to a row vector or a column vector, and then cascading to obtain a classification feature vector; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
The intelligent production method of the aluminum cup further comprises training the convolutional neural network model using a spatial attention mechanism and the classifier.
Fig. 4 is a flowchart for training the convolutional neural network model using the spatial attention mechanism and the classifier in the intelligent production method of aluminum cups according to the embodiment of the application. As shown in fig. 4, the training of the convolutional neural network model using the spatial attention mechanism and the classifier includes the steps of: s210, acquiring training data, wherein the training data comprises training detection images of the machined and formed aluminum cup and a true value of whether the punching of the machined and formed aluminum cup is deviated or not; s220, masking the corresponding position of the training detection image based on the position of the hole in the aluminum cup to obtain a training masking image; s230, aggregating the training mask image and the training detection image along a channel dimension to obtain a training multi-channel detection image; s240, passing the training multi-channel detection image through the convolutional neural network model using a spatial attention mechanism to obtain a training classification characteristic diagram; s250, passing the training classification characteristic diagram through the classifier to obtain a classification loss function value; and S260, training the convolutional neural network model using a spatial attention mechanism and the classifier based on the classification loss function value and through back propagation of gradient descent, wherein in each round of iteration of the training process, a free label optimization factor of the training classification feature graph based on cross-classifier soft similarity is calculated as a label value of the classifier.
In particular, in the technical solution of the present application, the mask image and the detection image are aggregated along a channel dimension to obtain the multi-channel detection image, so that high-dimensional image semantic features of the mask image and the detection image can be fully extracted, but at the same time, expression of the high-dimensional image semantic features of the mask image and the detection image in the classification feature map can reduce association degree of feature distribution of the whole classification feature map. Further, since the multi-channel detection image obtains the classification feature map through a convolutional neural network model using a spatial attention mechanism, the association of the high-dimensional spatial semantic feature distribution between the mask image and the detection image is further weakened under the condition of strengthening the respective high-dimensional spatial semantic feature distribution of the mask image and the detection image, so that the overall feature distribution of the classification feature map may have strong discreteness, and training of a classifier, particularly convergence of a label value of the classifier is difficult.
Thus, soft label learning is preferably used instead of usual hard label learning, in particular, at each iteration, a free label optimization factor of the classification feature map based on cross-classifier soft similarity is calculated as the label value of the classifier, expressed as:
Figure SMS_33
，
Wherein the method comprises the steps of
Figure SMS_36
Is the training classification feature vector obtained after the training classification feature map is unfolded, and is +.>
Figure SMS_37
Is a weight matrix of training classification feature vectors obtained by the classifier after the training classification feature map is developed, and is->
Figure SMS_41
And->
Figure SMS_35
Representing tensor multiplication and tensor addition, respectively,/->
Figure SMS_39
Representing the distance between vectors, < >>
Figure SMS_42
Representing the two norms of the vector, and +.>
Figure SMS_43
And->
Figure SMS_34
Is a weight superparameter,/->
Figure SMS_38
An exponential operation representing a vector representing a calculation of a natural exponential function value raised to a power by a characteristic value of each position in the vector, ">
Figure SMS_40
A tag value representing the classifier.
The free label optimization factor based on the cross-classifier soft similarity is used as a label value of the classifier, the classification probability of the classification feature map can be calculated instead of a hard label value, the classification feature vector obtained after the development of the classification feature map and the weight matrix of the classifier are subjected to bidirectional clustering, so that the pseudo class based on the weight matrix of the classifier is simulated through the soft similarity of the classification feature vector and the weight matrix, and therefore classification quantization loss caused by the hard label learning is avoided through soft similarity learning, free label optimization which is more focused on an internal weight structure of the classifier is realized, training of the label value of the classifier is optimized, and training speed of the classifier is improved. Therefore, the punching quality of the machined aluminum cup can be accurately detected, and the working efficiency and the production quality of the machining of the aluminum cup are improved.
In summary, an intelligent production method of aluminum cups based on the embodiments of the present application is illustrated, which adopts an artificial intelligent detection technology based on deep learning to perform feature mining on an interested region of a hole by anchoring a target at a position of the hole in a detection image, so as to extract hidden feature information about the hole with a small scale, and also utilizes the position feature information about the hole in the detection image to comprehensively detect whether the punching of the machined aluminum cup is offset. Therefore, the punching quality of the machined aluminum cup can be accurately detected, and the working efficiency and the production quality of the machining of the aluminum cup are improved.
Exemplary System: fig. 5 is a block diagram of an intelligent production system for aluminum cups according to an embodiment of the present application. As shown in fig. 5, an intelligent production system 100 for aluminum cups according to an embodiment of the present application includes: a detection image acquisition unit 110 for acquiring a detection image of the formed aluminum cup; a mask applying unit 120 for applying a mask to the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a mask image; a multi-channel aggregation unit 130, configured to aggregate the mask image and the detection image along a channel dimension to obtain a multi-channel detection image; a spatial attention applying unit 140 for applying the multi-channel detection image to a convolutional neural network model using a spatial attention mechanism to obtain a classification characteristic map; and a detection result generating unit 150, configured to pass the classification feature map through a classifier to obtain a classification result, where the classification result is used to indicate whether the punching of the machined aluminum cup is offset.
In one example, in the above-described intelligent production system 100 for aluminum cups, the mask applying unit 120 is configured to pass the detection image through a hole target recognition network to anchor the holes at the positions of the aluminum cups.
In one example, in the intelligent production system 100 of aluminum cups described above, the hole target identification network is Fast R-CNN, fast R-CNN, retinaNet.
In one example, in the above-described intelligent production system 100 for aluminum cups, the spatial attention applying unit 140 is further configured to: performing depth convolution encoding on the multi-channel detection image by using a convolution encoding part of the convolution neural network model to obtain an initial convolution characteristic diagram; inputting the initial convolution feature map into a spatial attention portion of the convolution neural network model to obtain a spatial attention map; -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile; and calculating the position-wise point multiplication of the spatial attention characteristic diagram and the initial convolution characteristic diagram to obtain a classification characteristic diagram.
In one example, in the above-mentioned intelligent production system 100 for aluminum cups, the detection result generating unit 150 is further configured to: expanding each classification feature matrix in the classification feature map into a one-dimensional feature vector according to a row vector or a column vector, and then cascading to obtain a classification feature vector; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
In one example, in the aluminum cup intelligent production system 100, a training module 200 for training the convolutional neural network model using the spatial attention mechanism and the classifier is further included.
FIG. 6 is a block diagram of a training module in an intelligent production system for aluminum cups according to an embodiment of the application. As shown in fig. 6, the training module 200 includes: a training data obtaining unit 210, configured to obtain training data, where the training data includes a training detection image of a machined aluminum cup, and a true value of whether a hole of the machined aluminum cup is shifted; a training mask applying unit 220, configured to apply a mask to the corresponding position of the training detection image based on the position of the hole in the aluminum cup to obtain a training mask image; a training multi-channel aggregation unit 230, configured to aggregate the training mask image and the training detection image along a channel dimension to obtain a training multi-channel detection image; a training spatial attention applying unit 240, configured to pass the training multichannel detection image through the convolutional neural network model using a spatial attention mechanism to obtain a training classification feature map; a classification loss unit 250, configured to pass the training classification feature map through the classifier to obtain a classification loss function value; and a training unit 260 for training the convolutional neural network model using a spatial attention mechanism and the classifier based on the classification loss function value and by back propagation of gradient descent, wherein, in each round of iteration of the training process, a free label optimization factor of the training classification feature map based on cross-classifier soft similarity is calculated as a label value of the classifier.
In one example, in the above-described aluminum cup intelligent production system 100, in each iteration of the training process, the free label optimization factor of the training classification feature vector based on cross-classifier soft similarity is calculated as the label value of the classifier with the following formula; wherein, the formula is:
Figure SMS_44
，
wherein the method comprises the steps of
Figure SMS_46
Is the training classification feature vector obtained after the training classification feature map is unfolded, and is +.>
Figure SMS_48
Is a weight matrix of training classification feature vectors obtained by the classifier after the training classification feature map is developed, and is->
Figure SMS_51
And->
Figure SMS_47
Representing tensor multiplication and tensor addition, respectively,/->
Figure SMS_50
Representing the distance between vectors, < >>
Figure SMS_53
Representing the two norms of the vector, and +.>
Figure SMS_54
And->
Figure SMS_45
Is a weight superparameter,/->
Figure SMS_49
An exponential operation representing a vector representing a natural exponential function exponentiated by the eigenvalues of each position in the vectorValue of->
Figure SMS_52
A tag value representing the classifier.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described intelligent production system 100 for aluminum cups have been described in detail in the above description of the intelligent production method for aluminum cups with reference to fig. 1 to 4, and thus, repetitive descriptions thereof will be omitted.
As described above, the intelligent production system 100 of aluminum cups according to the embodiments of the present application may be implemented in various terminal devices, such as a server for intelligent production of aluminum cups, and the like. In one example, the aluminum cup intelligent production system 100 according to embodiments of the present application may be integrated into a terminal device as a software module and/or hardware module. For example, the intelligent production system 100 of aluminum cups may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the aluminum cup intelligent production system 100 can also be one of a plurality of hardware modules of the terminal device.
Alternatively, in another example, the aluminum cup intelligent production system 100 and the terminal device may be separate devices, and the aluminum cup intelligent production system 100 may be connected to the terminal device through a wired and/or wireless network and transmit interactive information in a agreed data format.
Exemplary electronic device: next, an electronic apparatus according to an embodiment of the present application is described with reference to fig. 7. Fig. 7 is a block diagram of an electronic device according to an embodiment of the present application. As shown in fig. 7, the electronic device 10 includes one or more processors 11 and a memory 12.
The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
Memory 12 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like. On which one or more computer program instructions may be stored that the processor 11 may execute to implement the functions in the intelligent production method of aluminum cups and/or other desired functions of the various embodiments of the present application described above. Various contents such as a detected image of the machined aluminum cup may also be stored in the computer readable storage medium.
In one example, the electronic device 10 may further include: an input device 13 and an output device 14, which are interconnected by a bus system and/or other forms of connection mechanisms (not shown).
The input means 13 may comprise, for example, a keyboard, a mouse, etc.
The output device 14 may output various information including the classification result and the like to the outside. The output means 14 may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, etc.
Of course, only some of the components of the electronic device 10 that are relevant to the present application are shown in fig. 7 for simplicity, components such as buses, input/output interfaces, etc. are omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.
Exemplary computer program product and computer readable storage Medium: in addition to the methods and apparatus described above, embodiments of the present application may also be a computer program product comprising computer program instructions which, when executed by a processor, cause the processor to perform steps in the functions of the intelligent production method of aluminum cups according to the various embodiments of the present application described in the "exemplary methods" section of the present specification.
The computer program product may write program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present application may also be a computer-readable storage medium, having stored thereon computer program instructions, which when executed by a processor, cause the processor to perform steps in the functions of the intelligent production method of aluminum cups according to the various embodiments of the present application described in the above "exemplary methods" section of the present specification.
The computer readable storage medium may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The basic principles of the present application have been described above in connection with specific embodiments, however, it should be noted that the advantages, benefits, effects, etc. mentioned in the present application are merely examples and not limiting, and these advantages, benefits, effects, etc. are not to be considered as necessarily possessed by the various embodiments of the present application. Furthermore, the specific details disclosed herein are for purposes of illustration and understanding only, and are not intended to be limiting, as the application is not intended to be limited to the details disclosed herein as such.
The block diagrams of the devices, apparatuses, devices, systems referred to in this application are only illustrative examples and are not intended to require or imply that the connections, arrangements, configurations must be made in the manner shown in the block diagrams. As will be appreciated by one of skill in the art, the devices, apparatuses, devices, systems may be connected, arranged, configured in any manner. Words such as "including," "comprising," "having," and the like are words of openness and mean "including but not limited to," and are used interchangeably therewith. The terms "or" and "as used herein refer to and are used interchangeably with the term" and/or "unless the context clearly indicates otherwise. The term "such as" as used herein refers to, and is used interchangeably with, the phrase "such as, but not limited to.
It is also noted that in the apparatus, devices and methods of the present application, the components or steps may be disassembled and/or assembled. Such decomposition and/or recombination should be considered as equivalent to the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the application to the form disclosed herein. Although a number of example aspects and embodiments have been discussed above, a person of ordinary skill in the art will recognize certain variations, modifications, alterations, additions, and subcombinations thereof.

Claims (10)

1. An intelligent production method of an aluminum cup is characterized by comprising the following steps: acquiring a detection image of the machined and formed aluminum cup; applying a mask to the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a mask image; aggregating the mask image and the detection image along a channel dimension to obtain a multi-channel detection image; the multi-channel detection image is subjected to a convolutional neural network model using a spatial attention mechanism to obtain a classification characteristic diagram; and the classification characteristic diagram is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the punching of the machined and formed aluminum cup is deviated or not.
2. The intelligent production method of the aluminum cup according to claim 1, wherein the masking the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a mask image comprises: and enabling the detection image to pass through a hole target identification network so as to anchor the position of the hole in the aluminum cup.
3. The intelligent production method of the aluminum cup according to claim 2, wherein the hole target recognition network is Fast R-CNN or Fast R-CNN, retinaNet.
4. The method for intelligent production of aluminum cups according to claim 3, wherein said passing said multi-channel inspection image through a convolutional neural network model using a spatial attention mechanism to obtain a classification feature map comprises: performing depth convolution encoding on the multi-channel detection image by using a convolution encoding part of the convolution neural network model to obtain an initial convolution characteristic diagram; inputting the initial convolution feature map into a spatial attention portion of the convolution neural network model to obtain a spatial attention map; -passing said spatial attention map through a Softmax activation function to obtain a spatial attention profile; and calculating the position-wise point multiplication of the spatial attention characteristic diagram and the initial convolution characteristic diagram to obtain a classification characteristic diagram.
5. The method of intelligent production of aluminum cups according to claim 4, wherein said passing said classification feature map through a classifier to obtain classification results, said classification results being indicative of whether the perforations of the finished aluminum cups are offset, comprises: expanding each classification feature matrix in the classification feature map into a one-dimensional feature vector according to a row vector or a column vector, and then cascading to obtain a classification feature vector; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
6. The intelligent production method of the aluminum cup according to claim 1, further comprising a training step of: training the convolutional neural network model using a spatial attention mechanism and the classifier.
7. The intelligent production method of aluminum cups according to claim 6, wherein said training step comprises: acquiring training data, wherein the training data comprises training detection images of the machined and formed aluminum cup and a true value of whether the punching of the machined and formed aluminum cup is deviated or not; applying a mask to the corresponding position of the training detection image based on the position of the hole in the aluminum cup to obtain a training mask image; aggregating the training mask image and the training detection image along a channel dimension to obtain a training multi-channel detection image; passing the training multichannel detection image through the convolutional neural network model using a spatial attention mechanism to obtain a training classification characteristic diagram; passing the training classification feature map through the classifier to obtain a classification loss function value; and training the convolutional neural network model using a spatial attention mechanism and the classifier based on the classification loss function value and by back propagation of gradient descent, wherein, in each round of iteration of the training process, a free label optimization factor of the training classification feature graph based on cross-classifier soft similarity is calculated as a label value of the classifier.
8. The method for intelligent production of aluminum cups according to claim 7, wherein in each iteration of the training process, the free label optimization factor of the training classification feature vector based on cross-classifier soft similarity is calculated as the label value of the classifier with the following formula; wherein, the formula is:
Figure QLYQS_1
，
wherein the method comprises the steps of
Figure QLYQS_4
Is the training classification feature vector obtained after the training classification feature map is unfolded, and is +.>
Figure QLYQS_7
Is a weight matrix of training classification feature vectors obtained by the classifier after the training classification feature map is developed, and is->
Figure QLYQS_10
And->
Figure QLYQS_3
Representing tensor multiplication and tensor addition, respectively,/->
Figure QLYQS_6
Representing the distance between vectors, < >>
Figure QLYQS_9
Representing the two norms of the vector, and +.>
Figure QLYQS_11
And->
Figure QLYQS_2
Is a weight superparameter,/->
Figure QLYQS_5
An exponential operation representing a vector representing a calculation of a natural exponential function value raised to a power by a characteristic value of each position in the vector, ">
Figure QLYQS_8
A tag value representing the classifier.
9. An intelligent production system of aluminum cup, characterized by comprising: the detection image acquisition unit is used for acquiring a detection image of the machined and molded aluminum cup; a mask applying unit for applying a mask to the corresponding position of the detection image based on the position of the hole in the aluminum cup to obtain a mask image; a multi-channel aggregation unit, configured to aggregate the mask image and the detection image along a channel dimension to obtain a multi-channel detection image; a spatial attention applying unit for applying the multichannel detection image to a convolutional neural network model using a spatial attention mechanism to obtain a classification characteristic map; and the detection result generation unit is used for passing the classification characteristic diagram through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the punching of the machined and formed aluminum cup is deviated or not.
10. The intelligent production system of aluminum cups as recited in claim 9, wherein the mask applying unit is further configured to: and enabling the detection image to pass through a hole target identification network so as to anchor the position of the hole in the aluminum cup.
CN202310097580.XA 2023-02-10 2023-02-10 Intelligent production method and system for aluminum cup Withdrawn CN116167989A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310097580.XA CN116167989A (en) 2023-02-10 2023-02-10 Intelligent production method and system for aluminum cup

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310097580.XA CN116167989A (en) 2023-02-10 2023-02-10 Intelligent production method and system for aluminum cup

Publications (1)

Publication Number Publication Date
CN116167989A true CN116167989A (en) 2023-05-26

Family

ID=86412754

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310097580.XA Withdrawn CN116167989A (en) 2023-02-10 2023-02-10 Intelligent production method and system for aluminum cup

Country Status (1)

Country Link
CN (1) CN116167989A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116400426A (en) * 2023-06-06 2023-07-07 山东省煤田地质局第三勘探队 Electromagnetic method-based data survey system
CN116630909A (en) * 2023-06-16 2023-08-22 广东特视能智能科技有限公司 Unmanned intelligent monitoring system and method based on unmanned aerial vehicle
CN117078670A (en) * 2023-10-13 2023-11-17 深圳市永迦电子科技有限公司 Production control system of cloud photo frame
CN117291874A (en) * 2023-09-04 2023-12-26 深圳市众为精密科技有限公司 Automatic centering method and system for copper public measurement

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116400426A (en) * 2023-06-06 2023-07-07 山东省煤田地质局第三勘探队 Electromagnetic method-based data survey system
CN116400426B (en) * 2023-06-06 2023-08-29 山东省煤田地质局第三勘探队 Electromagnetic method-based data survey system
CN116630909A (en) * 2023-06-16 2023-08-22 广东特视能智能科技有限公司 Unmanned intelligent monitoring system and method based on unmanned aerial vehicle
CN116630909B (en) * 2023-06-16 2024-02-02 广东特视能智能科技有限公司 Unmanned intelligent monitoring system and method based on unmanned aerial vehicle
CN117291874A (en) * 2023-09-04 2023-12-26 深圳市众为精密科技有限公司 Automatic centering method and system for copper public measurement
CN117291874B (en) * 2023-09-04 2024-05-28 深圳市众为精密科技有限公司 Automatic centering method and system for copper public measurement
CN117078670A (en) * 2023-10-13 2023-11-17 深圳市永迦电子科技有限公司 Production control system of cloud photo frame
CN117078670B (en) * 2023-10-13 2024-01-26 深圳市永迦电子科技有限公司 Production control system of cloud photo frame

Similar Documents

Publication Publication Date Title
CN116167989A (en) Intelligent production method and system for aluminum cup
CN107766894B (en) Remote sensing image natural language generation method based on attention mechanism and deep learning
CN110209823B (en) Multi-label text classification method and system
CN109190524B (en) Human body action recognition method based on generation of confrontation network
CN110969020B (en) CNN and attention mechanism-based Chinese named entity identification method, system and medium
US20220382553A1 (en) Fine-grained image recognition method and apparatus using graph structure represented high-order relation discovery
CN102324047B (en) Hyper-spectral image ground object recognition method based on sparse kernel representation (SKR)
CN109766934B (en) Image target identification method based on depth Gabor network
CN116245513B (en) Automatic operation and maintenance system and method based on rule base
CN115951883B (en) Service component management system of distributed micro-service architecture and method thereof
CN112163114B (en) Image retrieval method based on feature fusion
CN115853173A (en) Building curtain wall for construction and installation
CN116091414A (en) Cardiovascular image recognition method and system based on deep learning
CN111291695B (en) Training method and recognition method for recognition model of personnel illegal behaviors and computer equipment
CN114299326A (en) Small sample classification method based on conversion network and self-supervision
CN116258947B (en) Industrial automatic processing method and system suitable for home customization industry
Verma et al. A hybrid K-mean clustering algorithm for prediction analysis
CN112465805A (en) Neural network training method for quality detection of steel bar stamping and bending
CN116467485B (en) Video image retrieval construction system and method thereof
CN116851856B (en) Pure waterline cutting processing technology and system thereof
CN116945258A (en) Die cutting machine control system and method thereof
US20200117838A1 (en) Method for generating a set of shape descriptors for a set of two or three dimensional geometric shapes
CN116678258A (en) Cold-heat exchanger for pressure vessel and control method thereof
CN116124448A (en) Fault diagnosis system and method for wind power gear box
CN116150371A (en) Asset repayment plan mass data processing method based on sharingJDBC

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20230526